Download as pdf or txt
Download as pdf or txt
You are on page 1of 2

Practical Programmer Robert L.

Glass

Searching for the Holy Grail


of Software Engineering
‘But my project is different’ really is a valid response.
“A strong method, like a specific size mic who said it has issued the ulti- problem-specific approaches with a
of wrench, is designed to fit and do mate insult. “Ad hoc,” in these aca- problem-independent facade. In
an optimal job on one kind of demic circles, is the ultimate dirty the discipline of method engineer-
problem; a weak method, like a word, the final “j’accuse” of deri- ing, for example, which has come
monkey wrench, is designed to adjust sion. around to the point of view that
to a multiplicity of problems, but Over the years some things have methodologies must be tailored to
solve none of them optimally.” begun to change in this regard. the problem domain at hand,
—Vessey and Glass, 1998 Now, there are more computer sci- there is almost always—in the lit-
ence academics who are willing to erature on the subject—someone

F
or decades now, computer consider the possibility that differ- who is proposing a
science academics have ent problems require different “meta” approach
been looking for the Holy solution approaches. to those tailored
Grail of software engineer- methodologies, as if
ing—the one true approach to there is still a generic
building software systems way out if we only apply sufficient
that can be applied, uni- brilliance.
versally, to any and all There has been, The term “ad hoc” itself is an
software projects. for example, a interesting case in point. As noted,
And for just as many positive response many computer scientists use it as
decades, software practi- from many readers of [4], a term of derision, interpreting the
tioners have looked at these which takes the position term to mean “unstructured” and
proposed Holy Grails and mut- that problem-focused solu- even “chaotic.” But the dictionary
tered something that sounds very tions are strong, and problem- has an entirely different definition
much like “...but our project is dif- independent solutions are weak, for the term. In all the dictionaries
ferent.” when viewed from the historic per- I referred to—and I have looked in
Those computer science acade- spective of the problem-solving many because the dictionary defin-
mics, of course, are not terribly discipline (this column’s introduc- ition I found is so different from
impressed by that response. They tory quote is taken from that arti- common computer science
sort of “nudge-nudge, wink-wink” cle). But still, as often as not, you usage—the term means “focused
at each other, as if to say “there go will find the same academics who on the problem at hand.” Oh,
those ignorant practitioners again, are willing to entertain the possi- really? That term of derision has
resisting what is good for them.” bility that different kinds of prob- thereby become an acceptable
And, as often as not, they mutter lems require different solution approach to problem solving, one
something about the practitioner approaches, returning like salmon our opening quote would even call
STEVE ADLER

use of the term “ad hoc solutions.” to their spawning area to the “strong.”
When you hear “ad hoc,” you notion of so-called “meta- All of these thoughts simmer
know the computer science acade- approaches;” collections of those beneath the surface of my many

COMMUNICATIONS OF THE ACM May 2002/Vol. 45, No. 5 15


Practical Programmer

When you hear the term “ad hoc,” you know the computer
science academic who said it has issued the ultimate insult. “Ad hoc,”
in these academic circles, is the ultimate dirty word, the final
“j’accuse” of derision.
beliefs about the computing field. of wonderful, insightful, human- understanding that “hook-up”
They account for my belief in soft- interest stories about many of the means different things in different
ware practice, for example, in the most critical issues of our day. businesses. One company observed
face of tremendous accusations Recently, in the face of booming its staff “has to be expert in linking
from some theorists who call it a interest in all the topics e-com- buyers and sellers in hundreds of
field in crisis. They account for my merce has spawned, the WSJ pub- categories,” and ties its success to
reversal of position on the field of lished a special section on the “its track record as a brick-and-
method engineering; I have always subject. And, among the many mortar firm” (it has built auto-
been dubious about what Tom topics covered in that section was mated tools, for example, that
DeMarco refers to as “Capital-M” an analysis of the high failure rate capture its traditional strength in
methodologies, on the grounds its of so many of those e-business “helping buyers buy the right
adherents insist users of the startups [1]. There were plenty of motor from the right specifica-
methodologies adhere to both the reasons given for the failures, the tions”).
letter and the spirit of their pro- ones you have probably read about There it is, once again. If you
scriptions. But now that I under- in other literature: poor business know how to provide solutions in
stand method engineering supports planning; focusing on vision more the computing field, you are only
the notion of tailored methodolo- than profit; management by techie; halfway to becoming a computing
gies—the kind DeMarco refers to failure to acknowledge the power expert. You also have to know how
with a small “m”—I can believe of the status quo; undercapitaliza- to address problems. And you need
the methodologies are beginning tion; overspending; unrealistic to understand—let me emphasize
to offer some real-world, useful expectations. These sort of things. the point—that solution approaches
concepts practitioners may be able (For a more thorough list, see [3].) must take into account the nature of
to put to good and (especially) cre- But at the end of that laundry the problem at hand. Even if some
ative use. list was another reason, one that would derisively call this way of
As I said, these thoughts often resonated with my own (very conducting business “ad hoc.” c
simmer beneath my surface. But biased) beliefs: failure to under-
every once in a while, they bubble stand the application domain of References
to the top. Here’s the latest exam- the underlying problem. Toy e-tail- 1. Anders, J. B2B: Yesterday’s darling. Wall
Street Journal (Oct. 23, 2000).
ple, an occurrence that caused me ers who don’t understand the low 2. Bannon, L. Toys: Rough play. Wall Street
to renew my rant on the subject of margins and advance purchasing Journal (Oct. 23, 2000).
3. Glass, R.L., Ed. ComputingFailure.com. Pren-
problem-focused solution requirements of the toys business tice-Hall, Englewood Cliffs, NJ, 2001.
approaches, the trigger that caused [2]. Healthcare e-sites that fail to 4. Vessey, I. and Glass, R.L. Strong vs. weak
me to write this column. understand the unique require- approaches to systems development. Com-
mun. ACM 41, 4 (Apr. 1998), 99–102.
The Wall Street Journal (WSJ ) ments of the medical profession
engages in some of my favorite (and believe me, there are plenty of Robert L. Glass (rlglass@acm.org) is the
journalism. Its editorial pages may them). B2B businesses that assume publisher of the Software Practitioner newsletter,
and editor emeritus of Elsevier’s Journal of Systems
reflect opinions somewhat to the their potential customers are ready and Software.
right of Genghis Khan on occa- for an electronic hook-up-buyers-
sion, but its news columns are full and-sellers approach, without © 2002 ACM 0002-0782/02/0500 $5.00

16 May 2002/Vol. 45, No. 5 COMMUNICATIONS OF THE ACM

You might also like